Sam Isaacson

@SA8416 what you did to keep the H20 amount constant is fine. You can also have a species within a reaction that stays constant if you do `kf, H20 -->...

There was a typo in my code it should be `f2! = of.f.f_iip`.

@yewalenikhil65 this is a big project as it will involve updates to ModelingToolkit's PDE support (both to handle these types of PDEs and to handle domain geometries), design changes here...

If you really want to work towards the PDE stuff, the first step is probably learning about the ModelingToolkit PDESystems and how they work / dispatch to DiffEqOperators and such.

@ChrisRackauckas comment is exactly what I was going to discuss. It seems like what is really needed is a well-documented representation for storing and getting information about a reaction network....

> Ok, I guess this discussion is getting out of hand: We actually have two topics, (1) qualitative analysis and (2) where to go with in-memory-models, IR and DSL. That's...

@Gaussia moving our discussion here as @ChrisRackauckas requested. For building a dependency graph; I think it would work best if there is a function to return the dependency graph given...

@Gaussia Great, thanks! If you've gotten too much to do between all these discussions I'd be happy to take charge of updating the `JumpProblem` construction to use both `MassActionRates` and...

@ChrisRackauckas Does Modeling Toolkit support generating ODEs with delays now? This might be a good reason to add MT output (at least start prototyping it).

I'm thinking that for now we can just have a separate method that gets applied to a reaction network to generate a MT model. As the simplification picture improves then...